Does Artificial Intelligence Prevail in Poverty Measurement?

Abstract

Artificial Intelligence (AI) has been used widely by many domains in academic research to explore and learn much ambiguity information from small to large dataset. It is also tremendously implemented in daily lives especially in late 20 centuries in diverse formation to enhance business scalability and improving business operation for better services and performances. This trend is also seen to evolve in the field of socioeconomic studies, with an individual or household economic and social status relative to the rest of society. Is this technology present in the field of socioeconomic especially in poverty measurement? What is the form of problem solved? Therefore, the authors try to answers these question through systematic review method from the existence of poverty measurement research until the beginning of 2019. A systematic literature search was performed in the Web of Science and Scopus to identify all potential relevant studies using Kitchenham, 2007 guideline. Of the 53 article documents, 15 papers were selected after subsequent title/abstract and full text screening related to poverty measurement. The findings show that Linear Regression is a popular method chosen and closely followed by Random Forest and Deep Learning. Most studies diversify the use of data sources to predict poverty more accurately. The tendency to use satellites data can be seen more significantly than other types of data. Overall from 2007 to early 2019, the potential for using AI in the socioeconomic remains open.